Arc Tan

MATH 675 HOMEWORK 1 SOLUTIONS TO SELECTED EXERCISES Exercise 3. Problem 1, p. 45, Kolmogorov. Let x, y, u, z X .
. (a) By the triangle inequality, (x, z) (x, y) + (y, z) (x, y) + (y, u) + (u, z). Therefore (x, z)(y, u) (x, y)+(z, u). A similar calculation yields (y, u) (x, y) (x, y) + (z, u). Hence |(x, z) (y, u)| (x, y) + (z, u) as desired. Alternately if one assumes that (b) then using the triangle inequality for the absolute value gives |(x, z) (y, u)| = |(x, z) (z, y) + (z, y) (y, u)| |(x, z) (z, y)| + |(z, y) (y, u)| (x, y) + (z, u) (b) By the triangle inequality, (x, z) (x, y) + (y, z) and (y, z) (y, x) + (x, z). Combining these inequalities gives (y, z) (x, z) (x, y) (x, z) (y, z) which is |(x, z) (y, z)| (x, y). Alternately if one assumes that (a) then (b) follows by taking z = u in (a). Exercise 4. Problem 2, p. 45, Kolmogorov.
1 2
(ai bj bi aj )2 = = = =
i=1 j=1
1 2 1 2
a2 b2 2ai bj bi aj + b2 a2 i j i j a2 b2 + i j
n n
i=1 j=1 n n
b2 a2 i j
ai bj bi aj
i=1 j=1 n n
i=1 j=1 1 n 2 n 2 a b 2 i=1 i j=1 j n n a2 b2 k k k=1 k=1
i=1 j=1 n n + b2 a2 i j i=1 j=1 n 2
ai bi
i=1 j=1
bj aj
ak bk
k=1
Since (ai bj bi aj )2 0 for all i, j,

n 2 n
ak bk
k=1
k=1
a2 k
n k=1
b2 k
which is the CauchySchwarz inequality.
Exercise 5. Consider the set D[0, 1] = {f : f is continuously dierentiable on [0, 1]}, and the function d(f, g) = maxx[0,1] |f (x) g (x)|. Show that d is not a metric for D[0, 1]. Can you think of any way to modify the set D[0, 1] so that d is a metric for this new set? d fails to be a metric because d(f, g) = 0 implies that maxx[0,1] |f (x)g (x)| = 0 which means that f (x) = g (x) for all x [0, 1]. This is true if and only if f and g dier by a constant. If this constant is nonzero then f = g, so that d(f, g) = 0 does not imply that f = g. There is more than one way to x this but what I had in mind is something like the following. Dene the equivalence relation on D[0, 1] by f g if f (x) = g (x) for all x [0, 1] or equivalently if f (x) g(x) is identically constant, and let D[0, 1] = {[f ]: f D[0, 1]} where [f ] is the equivalence class of f . A closely related solution is to let D[0, 1] = {f D[0, 1]: f (0) = 0} in which case D is a subset of D rather than strictly speaking a modication of it. One could also replace f (0) = 0 with f (a) = b for any xed a [0, 1] and b R. Exercise 6. Prove that
(1)n = . 4 n=0 2n + 1
(Hint: Use the Taylor expansion for arctan(x) to get a series representation for arctan(r) with 0 < r < 1. Then take the limit as r 1 . You must be explicit in justifying the exchange of this limit with the innite Taylor sum.) There are a few ways to do this. (a) One way is to simply cite Abels Theorem which states essentially that the Taylor series expansion is always valid at the endpoint of the interval of convergence as long as the function is continuous and the Taylor series converges at that point. It says in addition that convergence of the Taylor series to the function is uniform on the resulting closed interval. The precise statement can be found in any advanced calculus book. The result in question can be proved directly quite easily without resorting to this theorem. This is what I had in mind. (b) The Taylor series expansion for arctan(x) is
arctan(x) =
(1)n
n=0
x2n+1 , |x| < 1. 2n + 1
Citing the usual error estimate for alternating series (found in any calculus book) gives for every N and |r| < 1
N
arctan(r)
(1)n
n=0
r2n+1 = 2n + 1
(1)n
n=N +1
r2n+1 r2N +3 . < 2n + 1 2N + 3
Hence for 0 < r < 1 and any N

N (1)n 4 n=0 2n + 1
N r2n+1 arctan(r) + arctan(r) + (1)n 4 2n + 1 n=0
(1)n
n=0
N r2n+1 (1)n 2n + 1 n=0 2n + 1
r arctan(r) + + 4 2N + 3
2N +3
(1) (r2n+1 1) 2n + 1 n=0
Letting r 1 on the right side, the rst term goes to zero by the continuity of arctan, and the third term goes to zero because the sum is nite, and only the middle term remains giving the estimate
N (1)n 1 4 n=0 2n + 1 2N + 3
which implies the result. (c) The same estimate comes from the wellknown formula for the partial sum of a geometric series giving the formula
N 1 (t2 )N +1 = (t2 )n + 1 + t2 n=0 1 + t2
valid for every t R. Integrating both sides gives = arctan(1) 4 1 dt = 0 1 + t2

N
= = Therefore
(1)n
1 0
t2n dt + (1)N +1
1 0
1 0
n=0 N
t2N +2 dt 1 + t2
(1)n + (1)N +1 2n + 1 n=0
t2N +2 dt 1 + t2
N (1)n 4 n=0 2n + 1
1 0
t2N +2 dt 1 + t2
1 0
t2N +2 dt =
1 2N + 3
and the result follows as before.

Arc Tan

Uploaded by

Copyright:

Available Formats

Arc Tan

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Arc Tan

Uploaded by

Copyright:

Available Formats

MATH 675 HOMEWORK 1 SOLUTIONS TO SELECTED EXERCISES Exercise 3. Problem 1, p. 45, Kolmogorov. Let x, y, u, z X .

i=1 j=1 1 n 2 n 2 a b 2 i=1 i j=1 j n n a2 b2 k k k=1 k=1

i=1 j=1 n n + b2 a2 i j i=1 j=1 n 2

Since (ai bj bi aj )2 0 for all i, j,

which is the CauchySchwarz inequality.

x2n+1 , |x| < 1. 2n + 1

r2n+1 r2N +3 . < 2n + 1 2N + 3

Hence for 0 < r < 1 and any N

N r2n+1 arctan(r) + arctan(r) + (1)n 4 2n + 1 n=0

N r2n+1 (1)n 2n + 1 n=0 2n + 1

(1) (r2n+1 1) 2n + 1 n=0

valid for every t R. Integrating both sides gives = arctan(1) 4 1 dt = 0 1 + t2

(1)n + (1)N +1 2n + 1 n=0

and the result follows as before.

You might also like